MarjorieR

@moelassus@mastodon.social @markwyner@mas.to what we need is for the UK to adopt effective anti-SLAPP legislation. We need protection from the Billionaires. Now that would be something practical for Burnham to take up. https://antislapp.uk/solutions/legislation/

@urlyman@mastodon.social and now that the US military are being told to use 'AI' this happens: https://arstechnica.com/ai/2026/09/report-us-almost-boarded-chinese-ship-over-hallucinated-ai-arms-report/ It's a ticking time bomb.

@drahardja@sfba.social @librarysquirrel@sunny.garden they boast about the energy consumption - a proxy for size - of their datacentes because there is no other measurable metric of how big their ambition is. They certainly can't talk about it in terms is the notional output produced by GenAI as that is just slop and for vibe coding technical debt as it actually has negative value.

@lauerhahn@sfba.social @ChrisMayLA6@mastodon.me.uk transfer pricing, sometimes blatant but particularly for IP that isn't traded, has been a thing for ever. The evil oil companies were already doing this back in the 1960s and 1970s. I know because I had a uni. placement at one then.

@bodhipaksa@mastodon.scot given the lack of viable energy sources for this other proposed "AI" data centre in Lanarkshire, what are the odds that this one in Fife is also vapourware? https://www.theguardian.com/technology/2026/jul/06/lanarkshire-scotland-ai-datacentre-project-renewable-energy We can but hope that someone will check.

@juliette@mastodon.green and yes we have CCTV, to discourage and maybe catch customers and staff stealing from us, but not there.

@Social_Recluse2@mastodon.social @benroyce@mastodon.social well Orwell's 1984 comes to mind. In this particular context, his Ministry of Truth. Both GenAI (esp. Grok) and Trump busy rewriting history.

@Geri@veganism.social isn't it about time both Labour Together and the IDF be declared terrorist organisations?

@BaronessWinter@beige.party @kevinrns@mstdn.social @randahl@mastodon.social Asia Football Conferaton also join in and criticise Infantino's coup attempt. https://www.khaleejtimes.com/sports/football/asian-football-confederation-criticises-fifa-20-billion-world-cup-investment-plan

@TheTenuousOrder@meow.social @maxleibman@beige.party when the AI bubble finally bursts even NVidia will take a hit. Maybe they'll find some sort of market by making chips for devices that people use.

@emilymbender@dair-community.social @selea@social.linux.pizza any idea why I see this get this rather disturbing message for your QT thread about the ACM invitation to feedback? "This account has been hidden by the moderators of social.linux.pizza" Of course I can click "Show Anyway" but I've also seen this recently on another post (not by you) and I'm starting to worry what is prompting this attempted censorious behaviour by the social.linux.pizza moderators.

@CppGuy@infosec.space @orionkidder@writing.exchange @grammargirl@zirk.us the error rate also also depends on what is in the training data. That's no doubt is part of the problem with Grok, as it's training data contains many unreliable statements garnered from X as well as deliberate falsifications added. If the training data was just Wikipedia you would get more reliable results. For other AI vendors adding random chats from Facebook or Instagram or AI generated websites will also lower the accuracy. Claude Code may be slightly better, for now, because it is just plagiarising code. This won't last as the code repositories fill up with AI slop and these are flagged up as such and excluded by the crawlers. Likewise if you ask GenAI to summarise a document it may well incorporate data from its training data as well as the text you supply. The other reason is that GenAI just doesn't simply reproduce single sources, whatever their accuracy. It acts as a stochastic mixer: if you see a AI generated legal case reference some of it may come from one citation and some from another and the legal inference drawn maybe from something entirely different. Likewise if you ask GenAI to summarise a document it may well incorporate words from its training data as well as the text you supply.

@MostlyHarmless@thecanadian.social good guys (as well as many bad guys) wait until they are dead before someone else thinks they are good enough, rightly or wrongly, to be memorialised with a statue, gold or otherwise. Someone who arranges for a statue to be erected in their lifetime must be really insecure and also likely very aware that no one would erect one once they were gone.

@dalias@hachyderm.io @david_chisnall@infosec.exchange seems we are all too nerdy. According to this article by @davidgerard@circumstances.run (pivot-to-ai): "The goblins started showing up in GPT 5.1. OpenAI blames post-training, where you take an existing AI model and try to tweak the model’s outputs: training the model for the personality customization feature, in particular the Nerdy personality. We unknowingly gave particularly high rewards for metaphors with creatures." The “Nerdy” personality was retired — but the goblins leaked through to the rest of the GPT 5.5 model. It’s full of goblins." https://pivot-to-ai.com/2026/05/06/openai-chatgpt-goes-goblin-mode-let-none-say-model-collapse/ And then of course goblins start infesting their training data so LLM start chasing their own tail and model collapse begins.

@MollyBeth@toot.lgbt he does know there are still women* (mostly cis-), all with pretty low testosterone, in the US military? *even if they can't get promoted these days.

@2legged@mastodon.ie it was built on both. It was Welsh miners who dug out the coal for English landowners and South Wales was as much a colonial enterprise as the overseas British Empire. Just compare the imperial architecture in Cardiff Civic Centre with that in New Delhi.

@kibcol1049@mstdn.social I'm pretty sure that your analysis of what the Insurers, Cigna, are doing with their algorithm is entirely accurate but you can't infer from the 80% success rate of the 0.2% who do appeal as to what the success rate would be for the other 99.8% - you appeal when you know you have a very good case and a chance of actually winning. Also what proportion of all claims were rejected? Its not stated, just that 300k were reviewed and rejected. Of course an entirely unscrupulous company would simply deny all claims.

@mansr@society.oftrolls.com @FaithfullJohn@mastodon.scot @futurebird@sauropods.win @doomsdayrs@cyberpunk.lol Well they nearly all use just one of the two indexing engines. That actually good, assuming the indexing is pukka, as it restricts the amount of web crawling that gets done. It would be great if the LLMs could do the same as it would reduce the ridiculous amount for web traffic and load on servers they generate. Perhaps they could pay the internet archive to provide an API they could all use. The value of a search engine is how it uses that indexed data, and as we can see that differs wildly, with Google deliberately made their search bad so you spend more time on their site, or now get you to use their "AI summaries". Still while there are better alternatives, it would be nice if someone brought back structured search (meaningfull and, or, not, quoted text). Has anyone?

@vaurora@mstdn.social @effy@social.effy.space the image provides "Miss", and the Alt-Text "Mrs" as against "Mr" on the other end of the "scale". Surely both should be "Ms" to be properly woke (I use "Ms").

@ikuturso@mastodon.social @_elena@mastodon.social in exploring the relationship between #WSocial, #Bluesky and the AT protocol it's perhaps worth noting that Trumps Nazi Bar/megaphone Truth Social is actually based on a modified Mastodon platform. I don't think they go to any length to acknowledge that either.

@nazokiyoubinbou@mastodon.social @infobeautiful@vis.social much of this growth and wealth is illusory. We would be no less well off, in fact we would be better off, if the AI bubble were to burst now. I agree people with claims on notional AI 'assets ' including those of us with pensions in funds that have not been able to avoid being sucked in will lose wealth. But AI data centres and AI mind worms are societally counterproductive. There would be more resource, less pollution and more jobs available to do useful things if they were to simply disappear. At the moment most of us are living in the Tech Bros' AI Bezzle, the moment in time when the mark for what is in this case the AI fraud does not realise that they have been a victim, and may even feel that they have gained in the short term, only to being disillusioned later on.

@maxleibman@beige.party given that most mainstream economists, let alone members of the general public, don't understand economics it's hardly surprising that many are also racists.

@mbeddedDev@fosstodon.org @maxleibman@beige.party but nobody is paying much for it - yet. Although large, Chatbot revenues are pitiful compared to the cost of the investments their investors are making or planning to make in hardware and training and supplying AI and its hallucinations and psychoses. And when individuals and companies have to start paying realistic prices that cover costs and companies start to see the impact on their operations of all that technical debt they have created by vibecoding then demand for AI, instead of growing exponentially will actually drop and the AI companies share valuations with it.

@gorsefan@ravenation.club @noybeu@mastodon.social @finnmyrstad@eupolicy.social @uc@mastodon.scot just to clarify. I think with the Guardian if you subscribe then you are paying to support the publication and as a bonus if you pay them enough not see the adverts. I don't know what options they offer as to 'personalisation' of ads (aka tracking). AFAIK this is the same model as the streamers (including ITV and CH4) run, where it's cheaper if you accept adverts. In my case I run ad and tracker blockers and I neither see the adverts or pay to view. If I were to support them (I think I might if they didn't have so much fashy stuff as well as the good) I would still keep my blockers on.

@uc@mastodon.scot @finnmyrstad@eupolicy.social @noybeu@mastodon.social I think with the Guardian it's pay to avoid advertising, rather than pay to avoid being tracked? You can be advertised at without 'personalisation' and pay to avoid that. For myself I do neither, I use an adblocker.

@Quantillion@mstdn.io @Nonilex@masto.ai they have been learning from Netanahu: this is an Israeli Ceasefire - you cease, we fire.

@randulo@mastodon.social @0xabad1dea@infosec.exchange Is skimming a bad thing? I think its a skill if you are literate, not a deficit. But then I think the word I would use is "skim", rather than "scan". Particularly these days, with so much repetitive wordy AI slop, online I usually skim for interest and only read more closely if the text is actually interesting and insightful.

@TheBreadmonkey@beige.party notice the winner doesn't point to her team logo, throw her hands up in the air or fist pump as she crosses the line like a regular pro cyclist would.

"Pound and UK bonds stronger after Starmer pledges to stay on as PM

Government bonds rallied after Keir Starmer vowed to stay on despite heavy losses for Labour in the local elections, the Press Association reports."

• from the Guardian Politics live.

I'm always amazed at the lemmings -like behaviour of the City.

@PttP@mastodon.social I think there may be an issue in that the (weak version) ofrevealed preference is about whether someone chooses A over B subject to a budget constraint is a revealed preference. So is the person preferring to sell their kidney (A) to pay their rent over being unhoused (B) is also in a sense changing their budget constraint. Samuelson introduced the idea of revealed preference to replace the idea that people have utility functions and starting from observed behavior seems to me to be an improvement. I have much more problems with other theoretical economics axioms such as that people have perfect information when making their choices and that they are also perfectly consistent, particularly over time when other factors can then come into play. Disclaimer: I do have a first degree in Technological Economics but I studied it a long time ago (1969-73), so I'm both a a bit rusty but also an unreconstructed Neo-Keynesian.

@pvonhellermannn@mastodon.green I did get a sight of this. Awful. Same sickening playbook as the 'mainstream' media response to Polanski calling out the police for kicking a disturbed non-white human in the head.

@adamrice@c.im @LordWoolamaloo@mastodon.scot usually in the UK they do cycle. But sometimes the switch is broken (there are small indicators on the switch which then doesn't light up) and some of them can take a while so people just cross against the lights if they think it's clear. We also have 'Zebra' crossings that don't have lights and road users are expected to stop. But it helps if the pedestrian who is wanting to cross gives some indication that that is what they plan to do. I nearly ran down a guy with a pushchair who abruptly turned and started to cross the crossing without warning, eg stop and look, in front of me. As I was on a bike I was able to swerve round him but a car would have run them down. I need to be more careful in future and not assume.

@Natasha_Jay@tech.lgbt on the other hand do you think if you feed it enough text it will achieve AGI, or at least turn into Marvin the Paranoid Android?

@jwildeboer@social.wildeboer.net not entirely on hope: hope based on intermittent reinforcement; it sometimes seems to work.

@HI_Greens@mastodon.scot to be fair the demands we make on the government for pensioner income (they aren't actually doing all that well, particularity the poorest) and for NHS care are increasing as the population ages and becomes increasing badly fed and unfit. However we can afford to fund this if we chose to. Strictly (according to MMT) we don't need to increases taxes but it would be perhaps it would be easier to justify if we were to tax those best able to afford it more (the rich, particularly the tax-avoiding ultra-rich and multinationals such as Amazon) and to remove the profiteering private sector out of our national services - NHS, Care, Water, Railways, Energy.

@valcurious@lgbtqia.space @halfbyte@ruby.social my guess would have been polenta, but happy to be corrected. Though if it tasted of peas I'd say it looks more like pease pudding (mashed peas) than mushy peas. As to the chips (crisps) in England in shops they always seem to be offered as part of a ''meal deal" (sandwich, can of sugary fizzy pop and crisps) . So an absolute no-goer for me.

@RaffKarva@sunny.garden @ChrisMayLA6@zirk.us something of an omission then.

@leighms@mastodonapp.uk @DarkOptimism@mstdn.social @Silversnapples@mastodonapp.uk this link is not up to date, at least in its guidance as to how you do it from the NHS App. "- through the NHS App by clicking on "Your Health" and selecting "Choose if data from your health records is shared for research and planning"" It's actually under your Profile (icon)>Personal Details: Health Choices>Health data sharing decision. And you'll still need to confirm your Name, DoB and NHS number.

@iftas@mastodon.iftas.org surely setting up 2FA is good advice for everyone?

@joriki@infosec.exchange @Natasha_Jay@tech.lgbt National Security? The US government have already banned the release of the latest versions ( Anthropics's Claude Fable 5 and Mythos 5) because they are allegedly too dangerous to be out there in the wild. And more generally it's probably impossible to protect LLMs from malicious prompt injection. Prompt injection is how they work. https://www.bbc.co.uk/news/articles/c932g3v3e13o

@Geri@veganism.social even with that big rise an average of £958 that seems very cheap. Where I live in Teignmouth Band A is nearly twice that (£1807).

@Nonilex@masto.ai but has anyone in the US military actually briefed Trump or Hesketh or Trumps negotiators on the facts? I doubt anyone has had the guts, on pain of losing their job, to try and burst Trump's fact-free bubble.

@grantdenkinson@mastodonapp.uk from the Ofcom study: "Ofcom’s research measured mobile performance across 24 segments of key railway lines covering England, Scotland and Wales, making it the most comprehensive dedicated study available on the subject to date. It looked at how often a phone could achieve good performance, defined as minimum download speeds of 5 Mbits per second, upload speeds of at least 1.5 Mbits per second, and a response time of 50 milliseconds or less, which would typically let people make video calls, stream content or scroll social media". Coverage on Paddington to South Devon is very patchy but their standard seems a bit over demanding to me. I would be happy if I could load static webpages, send and receive texts and email and read Mastodon. I'd rather get that everywhere first than worry about bring able to stream TikToc or make video calls.

@pndc@social.treehouse.systems surely a Nazi Bar with asbestos in the walls and the furniture, the bar itself hollowed out by termites and the staff replaced by agents.

@ravenbait@mastodon.scot not all rice needs washing but yes just cooking it will take 15 minutes after the water is up to boiling. Can't check the linked recipe though as you need to register just to read it.

@PttP@mastodon.social "Under the doctrine of revealed preferences, someone who sells their kidney to make the rent has a revealed preference for only having one kidney". No, i think this misrepresent the concept. But clearly they have shown they do a revealed preference to live, with a roof over their head and something to eat, with one kidney than to be unhoused and starving with two. Of course it is capitalism, that means that there is a price attached to selling a kidney and that denies poor people the money to live and that forces them to make these insidious choices and just down the line the money they get for selling that kidney won't last that long and they will die.